One of the world’s top investors has backed a UK-headquartered AI company in a $50 million round of funding as it launches into the United States.
Partly was co-founded by CEO Levi Fawcett in New Zealand in 2020 and has built the world’s only AI foundation model purpose-built for the automotive repair supply chain.
Fawcett is a rocket scientist who managed and developed hardware simulations for the Electron Rocket at Rocket Lab USA before co-founding Partly. Following years of research, he launched the product into the UK market and relocated the firm’s headquarters to London.
“The UK was pretty much from day one our biggest market,” Fawcett tells BusinessCloud.
Partly’s Interpreter model is the result of five years of training on human feedback and synthetic data, more than 50 manufacturer agreements, and continuous training on live data.
It is “orders of magnitude” more accurate than large language models in identifying damage and the parts needed to fix it, says Fawcett – but he is grateful for the emergence of LLMs nonetheless.
“At Rocket Lab I was working on forefront machine learning technology – trying to predict anomalies and failures,” Fawcett explains. “It was pretty clear back in 2020 that something like ChatGPT would emerge: the scaling laws were clear to everyone in the industry.
“When ChatGPT launched, two years in, it was a good thing – because people got really excited about AI, and have only got more and more excited since.”
After the first versions of the product didn’t work as planned – with the firm still pre-revenue – Partly’s team “got to the point where it’s almost as good as a human parts interpreter”.
“We then got our first million dollars in revenue, just proving that it worked,” says Fawcett. “Then in March last year we trained another version of the model that was more accurate than even a veteran human parts interpreter – and that was where the business exploded.
“We’ve been growing at 10% a week for months… we went from 40 people to 160 people over 18 months.”
Now DST Global Partners – the investor behind industry-defining brands like Anthropic, Meta, Alibaba, Airbnb and Spotify – has led its Series B raise as it targets the $100 billion+ US collision repair market, which has so far operated without AI-native infrastructure and relied on analogue solutions.
“Not since the creation of the assembly line or EVs (electric vehicles) has the auto industry experienced significant innovation that simultaneously improves operational efficiency, industry profitability, and consumer value,” said Fawcett.
Fawcett moved out to the Bay Area of San Francisco late last year while Partly has anchored its US operations in Austin, Texas. He now splits his time between the two, while Partly maintains its UK base as well as two offices in New Zealand.
He says the model is surprisingly adaptable to the different vehicle types in the US, once the core job of labelling car parts – carried out by its 40-strong team of ex-parts interpreters – is completed.
“What we spend a lot of our time doing is collecting very reliable and accurate context. What exactly has happened here? What are the exact parts needed? What do we repair? What do we replace? Why was it done like that?
“You’ve then got your machine learning researchers who train the model to predict these gold examples.”
He adds: “We try to hire the top 0.1% of talent globally, and the technology is very, very good – which means that we can scale fast, we can raise fast, and we can make good money… and that means we can then hire really good sales and marketing people.
“I hire for my weaknesses, and then focus more of my time on the things that I’m uniquely good at.”
Fawcett said the fundraising was completed in just two weeks. “DST Global Partners has one of the best track records of any investor: they’ve only missed investing in two of the $100bn+ public companies out there.
“We raised the $50m to keep growing fast and train a bigger version of the model.”
Fawcett likes to get out on his dirt bike on his farm on Sundays – and spent time with his kids – but Saturdays are spent in a deep research dive as he ‘zooms out’ on the business to identify its needs.
Asked about his leadership style, he answers: “It’s all about the theory of constraints.
“All that matters is what’s on the critical path this week, and then what’s going to be on the critical path in the next six months.
“My job, as the founder and CEO, is to predict as accurately as possible what’s going to be on the critical path in six months: what are going to be the fundamental constraints of the business? I’ll then dig deep into the two or three things that are blocking the business.
“If anyone’s on the critical path, I expect 24/7 work – you are working till 2am in the morning every day… you work until you sleep, then you get up and you work some more until you’re off critical path.
“So every single person’s job is to stay off critical path.”